Erosion and Karst in Subsurface Middle Paleozoic Rocks in the Arkoma Basin, Oklahoma, USA

Abstract

Seismic attribute analysis, guided by well data, reveals widespread stratigraphic anomalies caused by erosion or karstification in the late Ordovician-early Devonian Hunton Group in the Arkoma Basin, eastern Oklahoma, USA. This study shows that these strata are more extensive than previously known. Well data and seismic mapping in the Red Oak petroleum field identify approximately 40 m thick Hunton lenses, averaging 3 km in diameter, surrounded by karsted rock. These lenses may be remnants of incomplete erosion during the Middle Devonian period (pre-Woodford unconformity) or result from Hunton rocks sagging into sinkholes caused by karstification and collapse of underlying Viola or Bromide carbonates. Using formation tops from wells, correlated with attribute and structure maps from a 3D seismic volume, we identify (1) areas lacking Hunton seismic markers, indicating complete removal; (2) areas with Hunton contacts, showing where Hunton remains; and (3) zones without Hunton but with a thin layer underlying carbonate strata, interpreted as an incipient karst zone, often near areas with Hunton. We also observe that the thickness of the overlying Woodford Shale, a key hydrocarbon target, correlates with karstic and erosional thinning of Hunton carbonates. Interpretation of 3D seismic data reveals a strong connection between thinned Hunton and thickened Woodford Shale.

1. Introduction

The Hunton Group is an Ordovician-Devonian heterolithic succession of carbonates and shales deposited on a south-dipping ramp [1,2] (Figure 1a,b). It is a significant historical oil and gas producer in Oklahoma (USA). Today, the overlying Woodford Shale is the most economically important drilling target of the Arkoma Basin [3], with over 2493 wells drilled and well over 2 trillion cubic feet of gas (TCFG) produced [4]. However, in the Arkoma Basin of eastern Oklahoma, relatively few wells have penetrated the Hunton Group, and little is known about its stratigraphic character and distribution [5]. It exists, mainly in the subsurface, from the Texas Panhandle across most of Oklahoma and into western Arkansas [1,6]. Oil was first discovered in the Hunton Group in 1921 in eastern Oklahoma [7]. Although current exploitation of the Hunton Formation is overshadowed by the horizontal tight hydrocarbon play in the overlying Woodford Shale, it has historically produced enormous quantities of oil and natural gas, especially in the Anadarko Basin of western Oklahoma. Between January 1977 and September 1999, the Hunton Group produced 157,118,000 barrels of oil and 46,759,666 million cubic feet of natural gas [7].

Previous Hunton research has primarily focused on surface geological mapping in historically significant producing regions, such as the southern Anadarko Basin and the Cherokee Platform [1,7]. The Hunton Formation in the southern part of the Arkoma Basin in eastern Oklahoma has attracted somewhat less attention because across much of the basin it is relatively thin (Figure 2). It is deeply buried, averaging about 12,000–16,000 ft (3658–4877 m) (Figure 3). In a broad zone within the middle of the basin, approximately 25 mi (40 km) wide and 100 mi (161 km) long (Figure 2), Hunton rocks were entirely eroded during the Middle Devonian period.

Many subcrop isopach maps of the Hunton have been produced over the past century [6,8,9,10,11]. All show Hunton rocks thinning southward and eventually pinching out in the central Arkoma Basin (Figure 2). Using new data and interpretations presented herein, we will conclude that Hunton rocks occur in the subsurface much further south than previously believed. New drilling has encountered Hunton rocks in areas where they were previously thought to be absent. Understanding these newly discovered Hunton trends has become economically significant with the ongoing development of the Woodford horizontal gas play.

2. Purpose of Study

Our main goal is to map erosional and karstic features using seismic attributes. An additional aim is to update existing knowledge about the regional extent and thickness of the Hunton Group in the Arkoma Basin. Hunton strata were mapped with seismic data to better understand stratigraphic issues related to the southern erosional margin of the Hunton. To do this, we identified the most effective methods to resolve and map the erosional edge through 2D and 3D seismic and well data. Seismic attribute analysis was used to distinguish Hunton strata from surrounding rocks. The variability of rocks directly above the Hunton—ranging from thin conglomerates to cherty shales and sandstones—required the use of 3D seismic attributes for visualization. Since the Hunton erosional pinch-out is not a simple feather edge but is dispersed into increasingly thinner lenses, it was also important to include the neighboring stratigraphy to model its depositional and preservation history. We mapped the Hunton on a 3D seismic volume within a producing gas field and modelled the coeval karstic and erosional processes involved in its preservation.

The 3D seismic survey used in this study originates from the Red Oak gas field, located within the triangle zone between the Ouachita Mountains and the southern Arkoma Basin [12]. The study area is approximately 4 mi (6.4 km) north of the Choctaw Fault, located in Latimer, Haskell, and Le Flore counties, Oklahoma, and is marked by the red box in Figure 2 and Figure 3. Hunton rocks are largely absent along most of the northern and eastern parts of the area. Where present, they consist of thin lenticular bodies of carbonate strata.

Our study focused on the interval from the top of the Woodford Shale to the base of the Viola Group, which lies beneath the Hunton Group and Sylvan Shale (Figure 4). This interval was selected because of the depositional relationship these rocks share with the Hunton. The base of the Sylvan Shale and the top of the Woodford Shale, which mark the lower and upper boundaries, respectively, form chronostratigraphic surfaces [13]. As a result, the unusual thicknesses of these units are inversely related: the Woodford Shale tends to be thickest where the Hunton Group is thin or absent, possibly indicating back-filled pre-Woodford channels or other mechanisms creating relief (e.g., karstification). Conversely, when there is little or no thickening of the Woodford over Hunton lenses, and when nearby areas show partially karsted Viola rocks, we suggest that karst-induced collapse of Viola strata caused Hunton rocks to be lowered below the paleo-regional base level, thus preserving them. We will demonstrate that a combination of these processes can occur: a Hunton lens may be structurally lowered due to karstification of the underlying carbonates, but because of its resistive nature, the Hunton lens forms a depositional high, leading the Woodford to thin over it.

3. Previous Studies

Seismic reflection data (2D and 3D) from the Red Oak petroleum field and other petroleum fields have been utilized for many years to examine the complex subsurface Paleozoic stratigraphy in the Arkoma Basin of eastern Oklahoma [14,15,16,17,18,19,20,21]. In particular, previous research has successfully applied seismic attribute analysis to the detailed structure of stratigraphic intervals, including the Hunton Group and the Woodford Shale [2,22]. The interpretation of karstic features in Paleozoic subsurface strata of the Arkoma Basin has been a primary focus of many of these investigations [2,19,22,23,24].

4. Geologic Setting

4.1. Structure and Tectonics

The Arkoma basin is a foreland basin, a gently compressed fold belt of Pennsylvanian and older strata that underlies much of eastern Oklahoma and western Arkansas, extending south to the Choctaw Fault, which experienced maximum displacement during the Early Permian Period [25] (Figure 1a,b). This fault marks the front of the Ouachita Mountains, which run along the southern part of the basin. The frontal zone lies along the Choctaw fault and consists of severely shortened, north- and northwest-vergent imbricated thrusts and tight folds, mainly of Pennsylvanian strata [26]. Generally, the basin becomes deeper and thicker towards the south, into the Choctaw Fault [27].

The Ouachita Mountains are part of a largely buried late Paleozoic fold-and-thrust belt that stretches from Alabama to northern Mexico. In Oklahoma, the Ouachita Mountains and the Arkoma basin can be divided into four tectonic provinces (Figure 1b) based on different structural styles [28,29]. From north to south, these are (1) the Arkoma foreland basin, (2) the frontal thrust zone, (3) the central thrust zone characterized by broad open synclines separated by tight, generally thrust-cored anticlines of mainly Mississippian and Lower Pennsylvanian turbidites, and (4) the Broken Bow uplift, comprising isoclinally folded and faulted Lower Ordovician to Lower Mississippian strata [30,31]. This study focuses on the central thrust zone, where the Red Oak 3D seismic volume is located.

The tectonic history of the Oklahoma part of the Ouachita orogen can be outlined as follows. During the late Precambrian and Early Cambrian, the opening of the Iapetus Ocean created an irregular, rifted margin of the North American craton [32]. Later, a broad, southward-dipping continental shelf developed along the southern edge of North America, transforming into a prolific carbonate ramp from the Late Cambrian to the Early Devonian [27]. These deposits consist of widespread, thin carbonate units separated by thin shale layers and unconformities. These layers are believed to represent periods of terrigenous input and sea level lowstands, as a fall in sea level would reactivate fluvial systems, causing a hiatus in carbonate deposition and providing energy to transport siliceous material. Continental convergence began in the Late Mississippian between North America and the Sabine plate along a southward-dipping subduction zone [25]. In the Atokan (middle Pennsylvanian) period, before the study area experienced extensive thrusting, the shelf was interrupted by predominantly down-to-the-south normal faulting, apparently caused by the obduction of the approaching Ouachita accretionary prism and the related loading of the continental margin [33,34]. The subsequent orogeny reached its peak in the Early Permian [25]. Regional tectonic transport within the Ouachita fold-and-thrust belt was directed north-northwestward, approximately perpendicular to the current orientations of fold axes and fault traces [30].

4.2. Stratigraphy and Depositional Environment

The Hunton Group comprises organo-detrital limestones, associated dolomites, and shales (Figure 4), which formed in a shallow sea on a gently south-dipping ramp during the Late Ordovician to Early Devonian period [35]. In the study area, its thickness ranges from non-existent to over 300 ft (91 m). Across the basin, Hunton strata generally dip south (Figure 3) towards the Choctaw Fault Zone; the deepest Hunton penetration in the Arkoma basin reaches approximately 20,000 ft (6096 m).

Hunton deposition was not continuous, with unconformities occurring in the Early Silurian, Middle Silurian, and Early Devonian [1]. Of the several major unconformities, the most significant for this project are the post-Upper Silurian Henryhouse pre-Lower Devonian Frisco erosional unconformity, defined as a third-order regional unconformity, and the post-Hunton, pre-Woodford erosional unconformity, classified as a second-order unconformity [36].

The post-Hunton, pre-Woodford erosional unconformity occurs in the Middle Devonian (Emsian-Givetian). The current extent of the Hunton is mainly influenced by erosion that occurred during the time of this unconformity (Figure 3) [1,37]. Because eustatic sea levels rose throughout the Devonian (Figure 4), this unconformity is generally regarded as indicating a period of significant uplift [1,8]. Some researchers believe that the unconformity instead marks a major lowstand of short duration, based on the lack of dolomite and karst features in the Frisco Formation, the uppermost unit of the Hunton Group (Figure 4) [38]. This study aims to provide evidence in support of the structural uplift hypothesis.

A significant transgression happened after the middle Devonian erosion phase, during which both the Misener Sandstone (the basal member of the Woodford, composed of a cherty, fluvial lag) and the Woodford Shale (Figure 4) were laid down. Dissolution of the pre-Woodford surface during this hiatus created a network of relatively deep fissures and caverns in many of the Ordovician-Devonian carbonates, some reaching 50 ft (15 m) across [39,40,41]. Where this unconformity did not remove the Hunton, its upper surface was often eroded by fluvial processes, ranging from very little erosion to complete removal. Earlier researchers have created isopach maps of the Misener member (Figure 4), which show a broadly down-to-the-south dendritic stream pattern, influenced by faulting and fracturing along the north–south Nemaha Ridge [11,38].

5. Methodology

This study commenced with a detailed stratigraphic analysis of the section, using a high-quality 3D seismic survey and supported by a comprehensive set of well data. The model and results from this initial analysis were then applied to a much larger dataset across the Arkoma Basin.

Two separate 3D seismic surveys were used for this study, recorded in 1999 and 2001. These surveys were merged during reprocessing to create the dataset analyzed here. Notably, the 2001 survey was, at the time, among the most densely sampled land datasets worldwide. It included over 1 million prestack traces per square mile (2.6 km²), and because of its 55 ft (16.8 m) common depth point bin size, its nominal fold of cover exceeded 700. The survey was designed with such density to mitigate strong surface wave contamination (ground roll), which affects much reflection data in the lower Pennsylvanian Red Oak Sandstone (the largest producing interval in the Red Oak gas field). Although the high fold did not entirely eliminate ground roll in the Pennsylvanian section, it is not a concern in the much deeper carbonate rocks involved in this study. Ultimately, this high nominal fold of cover provided high-quality data in areas of interest related to Hunton stratigraphy.

The combined migrated dataset includes 866 east–west in-line profiles and 1415 north–south cross-line profiles, totaling 1,225,390 bins (common cell gathers). The source line orientation was north–south, and the receiver line orientation was east–west in both surveys. The length in the cross-line direction is 12 mi (19.3 km), and in the in-line direction it is 8 mi (12.9 km), covering a total area of 103 square mi (266.8 km²). The data quality varies depending on fold coverage. Since we merged two datasets, the maximum fold in the center of the combined data was 288, decreasing towards the edges of the survey. The data were recorded in the field following SEG (Society of Exploration Geophysicists) standards, meaning that an upward earth motion results in negative amplitudes on the geophones.

We applied surface-consistent deconvolution, a process that eliminates undesired filtering by the Earth [42], assuming a minimum-phase source (i.e., dynamite). Special care had to be taken with phase consistency because this is a merged survey, due to the detailed stratigraphic interpretations derived from the waveform. To match the dynamite data, the vibroseis data were converted from zero phase (where the wavelet is symmetrical about zero time) to minimum phase (where the wavelet is causal, offset in time) before deconvolution. This sought to ensure consistency with the dynamite data and to enable a single deconvolution algorithm to be applied across the entire combined survey, resulting in zero phase data. Zero-phase data are preferred over minimum-phase data because they tend to offer sharper definition and less distortion between stratigraphic features in the subsurface [43], in this case aiding in the delineation of the Woodford-Hunton contact. Applying deconvolution ensures that the data are close to zero-phase. No other subsequent processing was applied that could alter the data’s phase. Further information on the acquisition and processing parameters can be found in [44].

We seismically mapped five horizons: the top and base of the Woodford Shale, the top and base of the Viola Group, and a weak, interrupted horizon within the Viola (Figure 5). Areas lacking Hunton were defined where the base of the Woodford Shale coincided with the top of the Viola Group. In zones containing Hunton, we subtracted the base of the Woodford from the top of the Viola to create a Hunton isochron map. The Sylvan Shale was not resolvable in our seismic data, so it was grouped with the Hunton. Travel time–structure maps were produced by mapping these horizons at every fifth vertical section, in both inline and crossline directions. Areas with Hunton rocks were mapped more precisely, from every third to every line, since their weak amplitudes made them difficult for automated interpolation algorithms (auto pickers) to track. Fault heaves, which are the horizontal offsets that separate once continuous formations at a normal fault, are shown as gaps in the horizons. Control polygon maps were created by connecting these gaps, indicating fault heaves. These maps form a series of polygons outlining each fault heave. The control maps were then used with an interpolation algorithm to construct a continuous horizon from the interpreted lines, ensuring the algorithm did not interpolate across faults to maintain accuracy. Finally, the horizons were used to generate accurate isochron maps of the Woodford, Hunton/Sylvan, and Viola intervals through subtractive calculations (Figure 6).

As we started to understand how these units connect, we chose the seismic attribute extractions that best demonstrate shared erosional and depositional relationships. Seismic attributes highlight features of the seismic response with geological importance, such as a high-amplitude fluvial channel contrasted against a low-amplitude background representing overbank muds [45]. The attribute extractions showed the zones of interest using three different methods: direct extraction via a total amplitude algorithm (Figure 7), a statistical analysis of the wavelet (to identify trends in the different frequency components of the waveform) over the zone of interest through a waveform extraction (Figure 8), and the isolation of wavelet components through frequency and phase extractions (Figure 8). Some extractions depicted Hunton rocks more clearly than others, including those that grouped Hunton rocks with unrelated deeper Viola stratigraphic variations. We validated the attribute extractions by comparing them with an isochron of the Hunton (Figure 6). We also tested several attributes, calculated through independent extracts, to validate the repeatability of our interpretations. Many attributes, such as the arc length extraction shown in Figure 8, accurately map the Hunton but tend to be grouped with the Viola stratigraphy (compare Figure 6 and Figure 8). Clumping of Hunton and Viola stratigraphy usually happened in frequency-dominated extractions, such as arc length, energy halftime, effective bandwidth, and dominant frequency series. The extractions used for this study are outlined in the Data Interpretation section.

6. Interpretation

6.1. Seismic Character of the Hunton

We began our interpretation by using synthetic seismograms from six wells to correlate formation tops with key stratigraphic horizons in the seismic sections. A notable first observation is the appearance of low-amplitude doublets (Figure 7) between the high-amplitude Woodford base and the top of the Viola Group, which is also a high-amplitude horizon. A well drilled in 2004 in the Red Oak gas field found wireline log characteristics of Hunton and underlying Sylvan Shale rocks, confirming that these low-amplitude doublets are Hunton rocks (Figure 7). Within the Hunton wedge perforated by the well, the base of the Woodford surface (top of the Hunton Group) was found to be 165 ft (50.3 m) above the Viola surface, separated by 102 ft (31.1 m) of Hunton rocks and another 63 ft (19.2 m) of Sylvan Shale (Figure 7).

We note that Figure 7 and other vertical views are displayed as “squash plots” where the traces are stacked close together to permit displaying a long section on a single page. This creates a vertically exaggerated image, which causes the interpreted stratigraphic “picks” to seem jagged. We note that the seismic picks are mostly on the zero-crossing and largely auto-picked to avoid subjective interpretations.

A synthetic seismogram was generated by first creating an acoustic impedance log from the density and sonic logs of our type well. This was then convolved with a source wavelet extracted from our seismic dataset. The resulting synthetic seismic trace helped define the seismic characteristics of the Woodford Shale, Hunton Group, and Viola Group, and was used to tie stratigraphic picks to the seismic data. The top of the Woodford Shale appears as a high-amplitude, continuous seismic trough, with the base following as a peak. Our synthetic model indicates this peak has a lower amplitude where Hunton rocks are present but is significantly higher when positioned over the Viola Group (Figure 7). The Hunton interval is characterized as a discontinuous, low-amplitude doublet, 20 to 30 ms in travel time thickness. The Sylvan Shale, with its lower seismic velocity, is indistinguishable from the Hunton, causing the base of the Hunton/Sylvan interval to appear as a single peak. This same high-amplitude, continuous peak also corresponds to the top of the Viola. The interpreted Viola horizon includes the underlying Bromide Dense Member of the Bromide Group, as they are seismically indistinguishable. The base of this interval is marked by the underlying high-amplitude, continuous trough (Figure 7).

The seismic response in each wedge of the Hunton varies significantly from the response where the Hunton is absent. This variation in seismic behavior is clearly demonstrated where the total amplitude of the interpreted Hunton–Viola wavelet is overlaid on the interpreted Viola time–structure map (Figure 7). Irregular amplitude lows align with the Hunton wedges, as shown by comparing the Viola surface (Figure 7) with the Hunton isochron map (Figure 6). This isochron surface follows a reflection trough between the interpreted base of the Woodford Shale and the top of the Viola Group.

6.2. Imaging Collapse Features Within the Viola Group

A notable feature of the Arkoma basin stratigraphy identified in this study is how Hunton rocks have been affected by solution collapse originating from deeper rocks of the Ordovician-age Viola Group, which has complicated the depositional thickness of the Woodford as it was laid down across the area. This is illustrated in a lithology log correlation in the Red Oak Gas Field (Figure 9 and Figure 10). Precise mapping of the Hunton wedges’ influence on Woodford thickening (Figure 6) and the shapes of these wedges (Figure 6) can only be achieved using 3D seismic data. Depressions are visible in the seismic-interpreted Viola surface (Figure 6). An important observation is that these Viola depressions are located directly beneath similar lenses of Hunton rocks approximately 100 ft (~30.5 m) thick, supporting the interpretation of a genetic link between the Hunton wedges and the underlying Viola depressions.

These collapse zones occur at a high spatial density, with adjacent collapses often separated by only 1 mi (1.61 km) or less (Figure 6). Due to the severe topographical disruption on the Hunton surface caused by the Viola collapses, they substantially affected Mississippian sedimentation in the overlying Woodford Shale. Such collapse features should therefore be considered when attempting to understand sedimentation patterns over karst-prone carbonates in general.

Event Similarity Prediction (ESP) attribute extractions essentially measure trace-to-trace dissimilarity across a set time window in the seismic data. The advantage of ESP dissimilarity maps (Figure 9) is that they reveal and highlight lateral seismic changes that may relate to geological anomalies. These maps provide visual identification of such features as faults, facies changes, and other geological patterns. Faults and stratigraphic changes (such as the collapse features within the Viola) often stand out in otherwise homogeneous displays [46]. One expects high trace-to-trace similarity where strata are flat and continuous, lower values when they are dipping but still continuous, and anomalous values when they are discontinuous. The ESP map (Figure 9) was calculated along the top of the Viola horizon and revealed strong circular and ovoid anomalies that correspond with mapped Hunton wedges. We interpret these anomalies as a truncation effect, meaning a line where continuous strata are cut by some geological feature. This karstening phenomenon is further illustrated by the interpreted sinkholes on an ESP map extracted from the Viola interval (Figure 9), from which the location, size, shape, and orientation of the Hunton lenses and interpreted sinkholes in Viola group rocks (near the Hunton base) can be derived. Since ESP volumes are essentially a measure of trace-to-trace similarity, discontinuities in the seismic data are enhanced [47]. The ESP map (Figure 9) shows that ovoid disruptions (white areas) occur across the top Viola horizon. Correlating this disruption pattern with the Hunton isochron map (Figure 6) confirms that each of these anomalies corresponds to a depression on the Viola surface.

The total amplitude and Viola ESP maps (Figure 7 and Figure 9) show that several depressions form a line nearly parallel to the normal fault system cutting through the Viola surface. This suggests that the same stress field responsible for normal faulting was active during the pre-Woodford unconformity. Fractures may have opened parallel to the fault lines, allowing meteoric waters to initiate dissolution. These depressions are typically oval, with the long axis parallel to the faults, ranging in length from approximately 500 ft (152 m) to 5000 ft (1524 m).

In the northern part of the Red Oak gas field, which we interpret as lacking Hunton wedges, the thickness of the Viola measured from well data is similar to that beneath the Hunton intervals. We suggest that the soluble part of the Viola responsible for karsting has been eroded (Welling Member?) (Figure 4) from both areas, leaving the more insoluble Viola Springs Member [48].

6.3. Using Amplitude Extractions to Characterize the Hunton Wedges

The intervals just above the interpreted sinkholes within the Viola exhibit a distinct seismic character interpreted as disrupted Hunton wedges. The behavior of the Hunton wedges is best illustrated when presented as an amplitude attribute overlay on a time–structure map of the Viola Group (Figure 7). Lateral amplitude variations can be used to distinguish areas of consistent stratigraphy from chaotic or mounded beds, e.g., [44]. Concordant beds show higher amplitudes because energy is not scattered. Hummocky beds display lower amplitudes, while chaotic beds exhibit the lowest amplitudes. To gain a clearer understanding of the Viola wavelet, we centered a very small travel-time window over the horizon (Figure 7). Focusing on a single wavelet enhances our interpretation of bed-to-bed interfaces. Summing the amplitude for all the samples on each trace within the interval reveals the oval and circular shapes of the Hunton lenses, with the negative amplitude anomaly confirming the chaotic nature of the collapsed rocks within the interpreted sinkhole. This low-amplitude character becomes larger and more clearly defined nearer to the edges of the collapsed intervals. We believe this results from the chaotic nature of the beds within the Hunton wedges. As the beds sagged, some brecciation and fracturing likely occurred, explaining the low amplitude. An important observation is that these low-amplitude anomalies occur in the thickest preserved sections of the Viola. This response is interpreted as incipient karsting, yet it had not developed sufficiently to cause the Hunton rocks to be lowered and thus preserved.

The location of profile ABCD shown in Figure 7 was chosen to cross four of the interpreted sinkholes on the Viola surface. A vertically exaggerated section (Figure 7) displays the consistently near-vertical attitude and the height of these stratigraphic disruptions. Each structural disruption starts in the Viola interval (just below 2 s) and extends up into the Hunton Group wedge. The vertical extent of these disrupted zones reaches up to 200 to 250 ft (61 m to 76 m) across the 3D seismic grid.

6.4. Using Waveform Classification to Differentiate Hunton from Viola Rocks

A six-part waveform classifier provides another example of how seismic reflection is sensitive to surface depressions within a travel-time window spanning from the base of the Woodford through the Viola (Figure 8). The waveform classifier allowed us to automatically group similar waveforms into classes [49]. These classes are ordered by similarity, with members of class 1 being more alike than those of class 2, and so on. In this study, six classes were chosen. We used an unsupervised algorithm to classify six waveform types, each representing different features within the search window (Figure 8), which we defined as the top of the Viola Group and the base of the Bromide Dense Formation. Three of these classes were particularly relevant to the Hunton wedges: one represented very thin to absent Viola rocks beneath Hunton wedges; another indicated a very thick Viola section with little or no karsting; and the third depicted Viola rocks that did not develop enough karsting to preserve a Hunton wedge. Hunton wedges are characterized by the thin Viola waveform class (Figure 8).

6.5. Modelling Hunton Preservation

An important observation is that the areas with the thinnest Woodford Shale are directly above the Hunton lenses, whereas thick moats of Woodford Shale surround the Hunton lenses. Further from the Hunton wedges, a more consistent Woodford thickness is observed. The “moats” are regions of unusually thick Woodford near the Hunton wedges and could be sinkholes where Hunton rocks have already been removed, creating an especially thick Woodford section (Figure 10 and Figure 11). Three stratigraphic models (Figure 11 and Figure 12) explain these variations: Model A (Figure 11) shows the typical Woodford case, while Models B and C (Figure 11) represent the thin and thick Woodford cases, respectively. The Woodford thickness over the Hunton wedges in the moat ranges from 250 ft (76 m) to 300 ft (91 m), suggesting that the Hunton rocks formed a paleo-high at least 50 ft (15 m) high, which was eventually flooded by Woodford seas (Model B, Figure 11). A seismic section beside an isochron of the Woodford (Figure 12) illustrates this thinning where the Woodford overlies a Hunton wedge and also shows the thinning of the Viola beneath the same Hunton wedge. We attribute the Hunton wedges to sinkholes that form in the underlying Viola, allowing the Hunton to sag as the Viola collapses, as depicted in Figure 6. The preservation of the Hunton wedges is primarily due to their location beneath the regional base level, as the area was eroded during the Middle Devonian period. However, since the Woodford thins over these wedges, we must consider that they still formed paleo-highs (Model B, Figure 11).

Our model, which shows Hunton wedges forming relative topographic highs during Woodford deposition, is best examined by noting the Sylvan Shale’s stratigraphic position relative to the base level. The Chimneyhill Subgroup of the Hunton Group generally resists erosion more than the weaker Sylvan Shale. The Sylvan Shale unit is stratigraphically sandwiched between the Hunton and Viola (Figure 4), sagging into sinkholes along with the Hunton during the Viola collapse. Non-karstified areas had their Sylvan sections above base level and thus were subject to erosion (Figure 13), typically accelerated by the removal of the weak Sylvan Shale and subsequent undermining and collapse of overlying Hunton rocks. Areas that sagged into sinkholes only had Hunton rocks exposed (Model B, Figure 11), with the Sylvan Shale buried within the sinkhole. Hunton carbonates, being more resistant, could stand up in relief much better than a typical Hunton-Sylvan-Viola section. The resulting Hunton wedges formed topographic highs as the Woodford Sea advanced and eventually drowned the region. The Woodford Shale is thus thinner over these Hunton paleo-highs, as demonstrated by a Woodford isochron map (Figure 12). The local thinning of the Woodford over Hunton wedges is an essential observation because this thinning could indicate areas of Hunton wedges otherwise not detected. We note that these slight variations in Woodford thickness are high-resolution features that do not appear on a basin-scale isopach map of the Woodford Shale (Figure 14).

6.6. Regional View

After conducting a detailed analysis of Hunton Group rocks in the Red Oak gas field (Figure 2) area, we extended our approach across the entire southern erosional margin of the Hunton in the Arkoma basin (Figure 1). We observed that these karsting episodes and the resulting wedges characterize the southern pre-Woodford erosional margin in at least two locations: the Red Oak field mentioned in this study and an area to the west on the borders of Pittsburg and Latimer counties (Figure 2). Further south, scattered well data demonstrate a gradual thickening of Hunton strata, which include not only the Chimneyhill Subgroup but also the Henryhouse and related strata. Since none of these wells indicate absent Hunton strata and all show a southward thickening of the Hunton, we conclude that the isolated Hunton wedges to the north become part of a continuous section of Hunton rock as we move down dip along the southern flank of a basement-involved arch that formed during the mid-Devonian (Figure 3).

Our regional isopach map of the Hunton Group features two extensive bodies of Hunton rocks separated by an east–west-aligned area approximately 20 mi (32 km) by 100 mi (161 km) where Hunton and, to a lesser extent, Sylvan rocks were eroded (Figure 2 and Figure 10). This geometry, aligned with the later Ouachita orogenic belt, suggests that a post-Hunton, pre-Woodford arch in mid-Devonian time may account for the broad region of thin or absent Hunton. Well-log correlations flattened on the top Woodford (Figure 10) indicate uplift along a broad, east–west oriented zone running the length of the Oklahoma part of the basin, with the Hunton eroded from the top of the arch but preserved on both limbs. The opposing dips (with the northern limb dipping to the north, and the southern limb dipping to the south) of Hunton and older rocks support a post-Hunton, pre-Woodford arch.

An isopach of the Woodford Shale (Figure 14) across the basin shows no influence from a pre-Woodford arch, indicating it was entirely levelled by the pre-Woodford unconformity. Rising eustatic sea levels (Figure 4) during the Devonian Period [50] support the presence of an arch, as the Arkoma Basin experienced a significant sea-level fall and subsequent erosion [1], pointing to structural uplift occurring alongside the Acadian Orogeny further northeast. West of the Arkoma Basin, other uplifts from the period after the Hunton and before the Woodford have been identified [8,51], with the Guthrie–Holdenville Arch being especially relevant to this study. The structure of regions where the Hunton formation is absent in the Arkoma Basin resembles an eroded arch, with its long axis aligned parallel to the later Ouachita contractional features.

7. Discussion

7.1. Geological Mechanism for Karst Collapse Structures

The vertical collapse zones are thought to originate from post-Viola carbonate dissolution beneath the Hunton, which occurred during the pre-Woodford exposure when the entire section was uplifted, possibly into the vadose zone (Figure 15). This hypothesis is supported by karst-derived vertical collapse zones seen in Viola outcrops in the Arbuckle Mountains, Garvin County (Figure 16) [41]. In fact, Viola karst exploration targets have been pursued throughout southern Oklahoma [52] (Figure 16).

The preservation of sediments through sagging into collapsed strata has previously been documented from outcrops [56,57] and other 3D seismic datasets [46,58]. Large karst-induced vertical collapse zones are also visible in outcrops of the Ordovician McKelligon Canyon Formation (roughly similar in age and composition to the Viola Group) in the Franklin Mountains near El Paso, Texas (Figure 16) [56,59]. The collapse features in the Franklin Mountains outcrop are very deep, with some extending vertically for at least 1200 ft (365 m) in the larger exposures. In Meade County, Kansas, subsurface lenses of St. Peter Sandstone age are believed to have been preserved in Ordovician-age sinkholes [60], through a mechanism similar to that described for the Hunton. These Kansas sinkholes formed in Arbuckle Group carbonates. Ref. [61] used 3D seismic data to document classic karst features [44] in the Gippsland Basin offshore South-East Australia, similar to what we interpret in the Oklahoma Arkoma Basin.

7.2. Potential for Solution Collapse Within the Viola

In a study of the Upper to Middle Ordovician Viola cores in south-central Oklahoma (Black Riverian to Richmondian), ref. [62] identified paleokarstic features such as collapse and crackle breccias, focused flow features, cave parabreccias, sediment infill, and solution-enhanced fractures (Figure 16). Solution-enlarged joints and bedding planes, karren, and small caves indicate active dissolution of the Viola. These features are common in outcrops of the Arbuckle Mountains, 25 mi (40 km) west of the study area. Paleokarst in the Viola cores and outcrops includes (1) cavern-fill parabreccia, (2) collapse breccia, (3) crackle breccia, (4) solution-enlarged fractures and vugs, sediment infill, and (5) conduits and channels [46,62,63,64]. All these features point to volumetrically destructive processes, supporting a model that would allow the Hunton to subside at the expense of the Viola.

We suggest that weathering during the formation of the Middle Devonian unconformity caused widespread Viola karstification and collapse in the study area, similar to what ref. [41] observed in south-central Oklahoma. Using a six-part waveform classifier (Figure 8), we correlate karst models (Figure 17) with different parts of the Viola Group interval, dividing the Viola into areas of early karst development, well-established karst, and regions where the Viola has been fully dissolved. The shapes of the interpreted collapse zones align with a karst terrain [64].

Although no constraining Viola cores are available within the project area, our seismic images allow a hypothesis regarding the origin of the collapse structures.

• Post-Viola/pre-Woodford fracture sets formed (trending east-northeast) in response to regional uplift during the Middle Devonian period.
• Karst solution weathering occurred between the post-Hunton and pre-Woodford periods, especially along vertical fractures trending northeast, where water seepage was intensified. This process created large caverns in some carbonate units, notably the Welling Member of the Viola Group.
• As uplift continued and the regional water table lowered, extensive Viola dissolution and collapse occurred.
• The resulting collapse structures caused Hunton rocks to fall below the paleo-regional base level, thereby protecting them from later widespread erosion.
• Episodes of collapse likely persisted until most of the solution caverns had caved in and filled from above, causing the overlying strata to sag.
• Due to the resistive nature of most of the Hunton Group, the collapsed Hunton lenses formed depositional highs as the Woodford Sea advanced and eventually submerged the region, causing the thinning of the resulting Woodford Shale over them.

8. Conclusions

Our 3D seismic reflection investigation, especially the use of attribute extractions, documents karstifying features and subtle stratigraphic changes that would be impossible to map using only 2D seismic reflection profiles. Clearly, our interpretation is most robust within the confines of the seismic dataset (Figure 1). Constraining the interpretation of 3D seismic data with previously unavailable well data shows the Hunton Group to be more extensive than previously believed. Mid-Devonian karsting episodes and resulting wedges typify the southern Hunton’s erosional edge. Similar wedges appear within and around the Red Oak and Wilburton fields. Further south, these isolated wedges become a continuous section of gradually thickening Hunton strata that ultimately include not only the Chimneyhill Subgroup but also the Henryhouse and related strata. Because these formations terminate in an erosional pinch-out and dip southwards with a source rock (the Woodford Shale) directly overlying them, they form ideal hydrocarbon targets.

We interpret the large zone of absent Hunton to the north of the Red Oak and Wilburton fields as the crest of a post-Hunton, pre-Woodford arch, based on the following: (1) Eustatic sea levels rose globally during the Devonian, but the southern mid-continent USA experienced a significant regression and erosion, indicating regional uplift. (2) West of the Arkoma basin, other post-Hunton, pre-Woodford arches have been identified, with the most significant to this study being the Guthrie–Holdenville Arch, which extends NW-SE through Hughes and Seminole Counties [8]. (3) The shape of the areas lacking Hunton in the Arkoma Basin corresponds with a breached arch (Figure 2).

We envisage a wide uplift that erodes the previously deposited Hunton rocks from the crest and, on the southern part of the arch, raises the Hunton through the Viola section into the vadose zone. This facilitates extensive solution weathering and karst development, ultimately shaping the Hunton wedge geometries observed today in the Red Oak and other nearby gas fields.